Publications by Year: 2013

Cigarette smoking is the major environmental risk factor for chronic obstructive pulmonary disease (COPD). Genome-wide association studies have provided compelling associations for three loci with COPD. In this study, we aimed to estimate direct, i.e., independent from smoking, and indirect effects of those loci on COPD development using mediation analysis. We included a total of 3,424 COPD cases and 1,872 unaffected controls with data on two smoking-related phenotypes: lifetime average smoking intensity and cumulative exposure to tobacco smoke (pack years). Our analysis revealed that effects of two linked variants (rs1051730 and rs8034191) in the AGPHD1/CHRNA3 cluster on COPD development are significantly, yet not entirely, mediated by the smoking-related phenotypes. Approximately 30% of the total effect of variants in the AGPHD1/CHRNA3 cluster on COPD development was mediated by pack years. Simultaneous analysis of modestly (r (2) = 0.21) linked markers in CHRNA3 and IREB2 revealed that an even larger ( 42%) proportion of the total effect of the CHRNA3 locus on COPD was mediated by pack years after adjustment for an IREB2 single nucleotide polymorphism. This study confirms the existence of direct effects of the AGPHD1/CHRNA3, IREB2, FAM13A and HHIP loci on COPD development. While the association of the AGPHD1/CHRNA3 locus with COPD is significantly mediated by smoking-related phenotypes, IREB2 appears to affect COPD independently of smoking.

Tract-based analysis from DTI has become a widely employed procedure to study the white matter of the brain and its alterations in neurological and neurosurgical pathologies. Automatic tractography selection methods, where a subset of detected tracts corresponding to a specific white matter structure are selected, are a key component of the DTI processing pipeline. Using automatic tractography selection, repeatable results free of intra and inter-expert variability can be obtained rapidly, without the need for cumbersome manual segmentation. Many of the current approaches for automatic tractography selection rely on a previous registration procedure using an atlas; hence, these methods are likely very sensitive to the accuracy of the registration. In this paper we show that the performance of the registration step is critical to the overall result. This effect can in turn affect the calculation of scalar parameters derived subsequently from the selected tracts and often used in clinical practice; we show that such errors may be comparable in magnitude to the subtle differences found in clinical studies to differentiate between healthy and pathological. As an alternative, we propose a tractography selection method based on the use of geometrical constraints specific for each fiber bundle. Our experimental results show that the approach proposed performs with increased robustness and accuracy with respect to other approaches in the literature, particularly in the presence of imperfect registration.

A Reply to the comment by Valerij G. Kiselev and Dmitry S. Novikov.

A Comment on the Letter by N. Shemesh, C.-F. Westin, and Y. Cohen, Phys. Rev. Lett. 108, 058103 (2012).. The authors of the Letter offer a Reply.

In this paper, we present a deformable registration framework for the diffusion tensor image (DTI) using polynomial expansion. The use of polynomial expansion in image registration has previously been shown to be beneficial due to fast convergence and high accuracy. However, earlier work was developed only for 3D scalar medical image registration. In this work, it is shown how polynomial expansion can be applied to DTI registration. A new measurement is proposed for DTI registration evaluation, which seems to be robust and sensitive in evaluating the result of DTI registration. We present the algorithms for DTI registration using polynomial expansion by the fractional anisotropy image, and an explicit tensor reorientation strategy is inherent to the registration process. Analytic transforms with high accuracy are derived from polynomial expansion and used for transforming the tensor’s orientation. Three measurements for DTI registration evaluation are presented and compared in experimental results. The experiments for algorithm validation are designed from simple affine deformation to nonlinear deformation cases, and the algorithms using polynomial expansion give a good performance in both cases. Inter-subject DTI registration results are presented showing the utility of the proposed method.

Accurate volumetric assessment in non-small cell lung cancer (NSCLC) is critical for adequately informing treatments. In this study we assessed the clinical relevance of a semiautomatic computed tomography (CT)-based segmentation method using the competitive region-growing based algorithm, implemented in the free and public available 3D-Slicer software platform. We compared the 3D-Slicer segmented volumes by three independent observers, who segmented the primary tumour of 20 NSCLC patients twice, to manual slice-by-slice delineations of five physicians. Furthermore, we compared all tumour contours to the macroscopic diameter of the tumour in pathology, considered as the "gold standard". The 3D-Slicer segmented volumes demonstrated high agreement (overlap fractions > 0.90), lower volume variability (p = 0.0003) and smaller uncertainty areas (p = 0.0002), compared to manual slice-by-slice delineations. Furthermore, 3D-Slicer segmentations showed a strong correlation to pathology (r = 0.89, 95%CI, 0.81-0.94). Our results show that semiautomatic 3D-Slicer segmentations can be used for accurate contouring and are more stable than manual delineations. Therefore, 3D-Slicer can be employed as a starting point for treatment decisions or for high-throughput data mining research, such as Radiomics, where manual delineating often represent a time-consuming bottleneck.

This paper investigates a diffeomorphic point-set registration based on non-stationary mixture models. The goal is to improve the non-linear registration of anatomical structures by representing each point as a general non-stationary kernel that provides information about the shape of that point. Our framework generalizes work done by others that use stationary models. We achieve this by integrating the shape at each point when calculating the point-set similarity and transforming it according to the calculated deformation. We also restrict the non-rigid transform to the space of symmetric diffeomorphisms. Our algorithm is validated in synthetic and human datasets in two different applications: fiber bundle and lung airways registration. Our results shows that non-stationary mixture models are superior to Gaussian mixture models and methods that do not take into account the shape of each point.

We propose a top-down fully automatic 3D vertebra segmentation algorithm using global shape-related as well as local appearance-related prior information. The former is brought into the system by a global statistical shape model built from annotated training data, i.e., annotated CT volumes. The latter is handled by a machine learning-based component, i.e., a boundary detector, providing a strong discriminative model for vertebra surface appearance by making use of local context-encoding features. This boundary detector, which is essentially a probabilistic boosting-tree classifier, is also learnt from annotated training data. Contextual information is taken into account by representing vertebra surface candidate voxels with high-dimensional vectors of 3D steerable features derived from the observed volume intensities. Our system does not only consider the body of the individual vertebrae but also the spinal processes. Before segmentation, the image parts depicting individual vertebrae are spatially normalized with respect to their bounding box information in terms of translation, orientation, and scale leading to more accurate results. We evaluate segmentation accuracy on 7 CT volumes each depicting 22 vertebrae. The results indicate a symmetric point-to-mesh surface error of 1.37 ± 0.37 mm, which matches the current state-of-the-art.

Amyloid deposits from several human diseases have been found to contain membrane lipids. Co-aggregation of lipids and amyloid proteins in amyloid aggregates, and the related extraction of lipids from cellular membranes, can influence structure and function in both the membrane and the formed amyloid deposit. Co-aggregation can therefore have important implications for the pathological consequences of amyloid formation. Still, very little is known about the mechanism behind co-aggregation and molecular structure in the formed aggregates. To address this, we study in vitro co-aggregation by incubating phospholipid model membranes with the Parkinson’s disease-associated protein, α-synuclein, in monomeric form. After aggregation, we find spontaneous uptake of phospholipids from anionic model membranes into the amyloid fibrils. Phospholipid quantification, polarization transfer solid-state NMR and cryo-TEM together reveal co-aggregation of phospholipids and α-synuclein in a saturable manner with a strong dependence on lipid composition. At low lipid to protein ratios, there is a close association of phospholipids to the fibril structure, which is apparent from reduced phospholipid mobility and morphological changes in fibril bundling. At higher lipid to protein ratios, additional vesicles adsorb along the fibrils. While interactions between lipids and amyloid-protein are generally discussed within the perspective of different protein species adsorbing to and perturbing the lipid membrane, the current work reveals amyloid formation in the presence of lipids as a co-aggregation process. The interaction leads to the formation of lipid-protein co-aggregates with distinct structure, dynamics and morphology compared to assemblies formed by either lipid or protein alone.